April 2, 2019

Data science

  • extracting knowledge and meaning from (big) data
  • statistics, mathematics, computer science


  • Where do the data come from?

(James Montgomery Flagg)

\(>\) 90% of researchers in the biological sciences work with or plan to work with big data

(Williams & Teal 2017)

Next-generation sequencing

(NIH National Human Genome Research Institute)

Next-generation sequencing

(NIH National Human Genome Research Institute)

Next-generation sequencing

(NIH National Human Genome Research Institute)

\(>\) 60% of researchers in the biological sciences report a need for more training in data science

Meta-analysis 2013 - 2016
(Attwood et al 2017)

Not just academia

We need to teach
data science in undergraduate life
science curriculum.

Barriers to data science integration

  1. Faculty training
  2. Student interest
  3. Student preparation in mathematics, statistics, and computer science
  4. Already overly full curricula
  5. Limited access to resources (hardware, software)

(Williams et al 2017)





Experiential
Data science for
Undergraduate
Cross-disciplinary
Education




Our goal

Modular integration of
data science curriculum into
existing courses

Content overview

Course overview

Students impacted per year

Example student

MICB 301

MICB 301 - 322

MICB 301 - 322 - 405

MICB 301 - 322 - 405 - 425

Example student

MICB 301

MICB 301

MICB 405

MICB 405

MICB 322

MICB 322

MICB 425

MICB 425

Solutions to integration

  1. Faculty training
  • Dedicated Postdoctoral Teaching and Learning Fellow
  • Cross-disciplinary TAs from multiple departments


2. Student interest

  • Direct connections to other course curricula
  • Hands-on, experiential learning

Solutions to integration

3. Student preparation

  • No prior knowledge assumed


4. Already overly full curricula

  • No new courses required


5. Limited access to resources

  • Stripped down datasets and use of cloud resources
  • Open-source tools and curricula

Does EDUCE effectively teach data science skills
to M&I students?

MICB 301
as a case study

EDUCE in MICB 301

  • 5 x 50 min class sessions across 5 weeks
  • Weekly assignments and a final report


  • Introduction to
    • data science
    • R/RStudio
    • statistics
  • Simple plots and running a t-test in R

Increased interest in data science

How would you rate your interest in…


Increased interest in data science

How would you rate your interest in…


Except…

No significant changes in interest in statistics

Increased experience in data science

What level of experience do you have in …


Increased experience in data science

What level of experience do you have in …


Except…

No significant changes in experience in statistics

Conclusions

  • Data science literacy is needed in the life sciences
  • EDUCE provides a flexible, modular approach for integrating data science into undergraduate curriculum
  • Even minimal exposure (5 hours) can increase student self-reported interest and experience in data science areas

The future

  • A wealth of survey data to mine

  • Repetition across 3 years for statistical analyses

  • More courses? Other departments?

  • Faculty of Science Data Science Committee

Acknowledgements

Steven Hallam
Jennifer Bonderoff

EDUCE TAs

Yue Liu (App MATH)
Julia Beni (U. Minnesota)
Kris Hong (CPSC, STAT)
Jonah Lin (MICB, CPSC)
Lisa McEwen (MedGen)
Ryan McLaughlin (BINFO)
Connor Morgan-Lang (BINFO)
Nolan Shelley (Botany)
David Yin (CPSC, STAT)

Course instructors

Sean Crowe
Lindsay Eltis
Jennifer Gardy
Marcia Graves
Martin Hirst
Bill Mohn
Dave Oliver
Jen Sibley

Collaborators

Gaby Cohen-Freue (STAT)
Patrick Walls (MATH)
Biljana Stojkova (ASDa)

Funding

UBC Teaching and Learning Enhancement Fund (TLEF)

NSERC CREATE Program (ECOSCOPE)

Department of Microbiology & Immunology

UBC Skylight and the Center for Teaching, Learning and Technology (CTLT)

Opportunities at UBC

References

Attwood TK et al 2017. A global perspective on evolving bioinformatics and data science training needs. Brief Bioinform. 20(2):398-404. doi: 10.1093/bib/bbx100

Williams JJ et al 2017. Barriers to integration of bioinformatics into undergraduate life sciences education. BioRxiv. doi: 10.1101/204420

Williams JJ & Teal TK. 2017. A vision for collaborative training infrastructure for bioinformatics. Ann N Y Acad Sci. 1387(1):54-60_ doi: 10.1111/nyas.13207

GenBank sequences

Undergraduate programs

BSc in Bioinformatics

  • U. of Montreal
  • U. Saskatchewan
  • U. Calgary
  • Carleton U.

Joint BSc degrees

  • Simon Fraser U.
  • U. of British Columbia

Specializations / minors

  • Dalhousie U.
  • McGill U.
  • U. of Toronto
  • U. of Victoria
  • U. of Waterloo
  • U. of Western Ontario

MDS programs

(Michael Rappa, NC State University)

Some prior experience

Minimal prior knowledge

Have you heard the term 'data science'?